regXwild

⏱ Superfast ^Advanced wildcards++? *,|,?,^,$,+,#,>,++??,##??,>c in addition to slow regex engines and more.

✔ regex-like quantifiers, amazing meta symbols, and speed...

Unique algorithms that was implemented on native unmanaged C++ but easily accessible in .NET through Conari (recommended due to caching of 0x29 opcodes + related optimizations), and others such as python etc.

Why regXwild ?

It was designed to be faster than just fast for features that usually go beyond the typical wildcards. Seriously, We love regex, I love, You love; 2013 far behind but regXwild still relevant for speed and powerful wildcards-like features, such as ##?? (which means 2 or 4) ...

🔍 Easy to start

Unmanaged native C++ or managed .NET project. It doesn't matter, just use it:

C++

#include <regXwild.h>
using namespace net::r_eg::regXwild;
...
EssRxW rxw;
if(rxw.match(_T("regXwild"), _T("reg?wild"))) {
    // ...
}

C# if Conari

using dynamic l = new ConariX("regXwild.dll");
...
if(l.match<bool>("regXwild", "reg?wild")) {
    // ...
}

🏄 Amazing meta symbols

ESS version (advanced EXT version)

EXT version (more simplified than ESS)

🧮 Quantifiers

1.3+ ++??; ##??

and similar ...

Play with our actual Unit-Tests.

🚀 Awesome speed

• ~2000 times faster when C++.
• For .NET (including modern .NET Core), Conari provides optional caching of 0x29 opcodes (Calli) and more to get similar to C++ result as possible.

Match result and Replacements

1.4+

EssRxW::MatchResult m;
rxw.match
(
    _T("number = '8888'; //TODO: up"),
    _T("'+'"),
    EssRxW::EngineOptions::F_MATCH_RESULT,
    &m
);
//m.start = 9
//m.end = 15
...
input.replace(m.start, m.end - m.start, _T("'9777'"));

tstring str = _T("year = 2021; dd = 17;");
...
if(rxw.replace(str, _T(" ##;"), _T(" 00;"))) {
    // year = 2021; dd = 00;
}

🍰 Open and Free

Open Source project; MIT License, Enjoy 🎉

License

The MIT License (MIT)

Copyright (c) 2013-2021  Denis Kuzmin <[email protected]> github/3F

[ ☕ Make a donation ]

regXwild contributors: https://github.com/3F/regXwild/graphs/contributors

We're waiting for your awesome contributions!

Speed

Procedure of testing

• Use the algo subproject as tester of the main algorithms (Release cfg - x32 & x64)
• In general, calculation is simple and uses average as i = (t2 - t1); (sum(i) / n) where:
  • i - one iteration for searching by filter. Represents the delta of time t2 - t1
  • n - the number of repeats of the matching to get average.

e.g.:

{
    Meter meter;
    int results = 0;

    for(int total = 0; total < average; ++total)
    {
        meter.start();
        for(int i = 0; i < iterations; ++i)
        {
            if((alg.*method)(data, filter)) {
                //...
            }
        }
        results += meter.delta();
    }

    TRACE((results / average) << "ms");
}

for regex results it also prepares additional basic_regex from filter, but of course, only one for all iterations:

meter.start();

auto rfilter = tregex(
    filter,
    regex_constants::icase | regex_constants::optimize
);

results += meter.delta();
...

Please note:

• +icase means ignore case sensitivity when matching the filter(pattern) within the searched string, i.e. ignoreCase = true. Without this, everything will be much faster of course. That is, icase always adds complexity.
• Below, MultiByte can be faster than Unicode (for the same platform and the same way of module use) but it depends on specific architecture and can be about ~2 times faster when native C++, and about ~4 times faster when .NET + Conari and related.
• The results below can be different on different machines. You need only look at the difference (in milliseconds) between algorithms for a specific target.
• To calculate the data, as in the table below, you need execute algo.exe

Sample of speed for Unicode

340 Unicode Symbols and 10^4 iterations (340 x 10000); Filter: L"nime**haru*02*Magica"

ESS vs EXT

350 Unicode Symbols and 10^4 iterations (350 x 10000);

For .NET users through Conari engine:

Same test Data & Filter: 10^4 iterations

Release cfg; x32 or x64 regXwild (Unicode)

Attention: For more speed you need upgrading to Conari 1.3 or higher !

How to get regXwild

regXwild v1.1+ can also be installed through NuGet same for both unmanaged and managed projects.

For .NET it will put x32 & x64 regXwild into $(TargetDir). Use it with your .net modules through Conari and so on.

x64 + x32 Unicode + MultiByte modules;

Please note: Modern regXwild packages will no longer be distributed together with Conari. Please consider to use it separately, Conari nuget packages.

• regXwild NuGet:
• GetNuTool: msbuild gnt.core /p:ngpackages="regXwild" or gnt /p:ngpackages="regXwild"
• GitHub Releases [ latest ]
• 🎲 CI builds: CI /artifacts ( old CI )